Apr 16 2024 61 mins 4
In episode 98 of the [i3] Podcast, we are speaking with Will Liang, who is an executive director at MA Financial Group, but is also well-known for his time with Macquarie Group, where he worked for more than a decade, including as Head of Technology for Macquarie Capital Australia and New Zealand. We discuss the application of AI in financial services and wealth management, ChatGPT and how to deal with AI hallucinations.
Overview of Podcast with Will Liang
02:30 When I was young I contemplated becoming a professional Go player
05:00 2016 was a life shattering moment for me; Lee Sedol was defeated by AlphaGo
07:00 I think generative AI will be a net positive for society
08:30 The impact of AI on industries will not be equally distributed
15:00 Brainstorming with ChatGPT or Claude
16:00 AI might help us communicate better
19:00 AI hallucinations are actually a fixable problem
22:30 Myths and misconceptions in AI
27:00 Most of the time when ChatGPT doesn’t work is because we are prompting it in the wrong way
28:30 Thinking Fast & Slow; AI is not good at thinking slow
29:00 Losing our jobs to AI? It is important to distinguish between the automation of tasks versus the automation of jobs
35:00 When implementing AI, look at where your data is and try to bring your application closer to the data
39:00 Don’t trust any third party large language model, instead deploy an open source model into your own cloud environment
43:00 You ask ChatGPT 10 times the same question and it will give you nine different answers. That is a problem.
45:00 Deep fake is a real problem
50:00 Future trends: AI agents
53:00 Generative AI will be more of a game changer for private markets than public markets
Overview of Podcast with Will Liang
02:30 When I was young I contemplated becoming a professional Go player
05:00 2016 was a life shattering moment for me; Lee Sedol was defeated by AlphaGo
07:00 I think generative AI will be a net positive for society
08:30 The impact of AI on industries will not be equally distributed
15:00 Brainstorming with ChatGPT or Claude
16:00 AI might help us communicate better
19:00 AI hallucinations are actually a fixable problem
22:30 Myths and misconceptions in AI
27:00 Most of the time when ChatGPT doesn’t work is because we are prompting it in the wrong way
28:30 Thinking Fast & Slow; AI is not good at thinking slow
29:00 Losing our jobs to AI? It is important to distinguish between the automation of tasks versus the automation of jobs
35:00 When implementing AI, look at where your data is and try to bring your application closer to the data
39:00 Don’t trust any third party large language model, instead deploy an open source model into your own cloud environment
43:00 You ask ChatGPT 10 times the same question and it will give you nine different answers. That is a problem.
45:00 Deep fake is a real problem
50:00 Future trends: AI agents
53:00 Generative AI will be more of a game changer for private markets than public markets
